Professional Documents
Culture Documents
Psdtheory
Psdtheory
The question is: what is the PSD? What does it mean? What is a “spectral
density,” and why is Sx called a power spectral density?
Z T
(4)
2
= x(t) e−2πif t dt
− T2
1
This restriction is necessary because not all of our signals will be square integrable.
However, they will be mean square integrable, which is what we will take advantage of
here.
c
2005 by J. P. Cusumano This revision: February 10, 2005.
and so
"Z T
# "Z T
#
2 2
XT XT∗ = x(t) e−2πif t dt x∗ (s) e2πif s ds
− T2 − T2
(5)
Z T Z T
2 2
= x(t) x(s) e−2πif (t−s) dt ds ,
− T2 − T2
where the star denotes complex conjugation and for compactness the fre-
quency argument of XT has been suppressed. Taking the expectation of
both sides of Eq. (5)2
Z T Z T
2 2
E[XT XT∗ ] = E[x(t) x(s)] e−2πif (t−s) dt ds . (6)
− T2 − T2
To actually evaluate the above integral, the both variables of integration must
be changed. Let
Then, the integral of Eq. (7) is transformed (except for the limits of integra-
tion) using the change of variables formula:3
Z T Z T Z Z
2 2
−2πif τ
Rx (τ ) e dt ds = Rx (τ ) e−2πif τ |J|−1 dη dτ , (9)
− T2 − T2
2
To understand what this means, remember that Eq. (5) holds for any x(t). So imagine
computing Eq. (6) for different x(t) obtained from different experiments on the same
system (each one of these is called a sample function). The expectation is over all possible
sample functions. Since the exponential kernel inside the integral of Eq. (6) is the same
for each sample function, it can be pulled outside of the expectation.
3
This is a basic result from multivariable calculus. See, for example, I.S. Sokolnikoff and
R.M. Redheffer, Mathematics of Physics and Modern Engineering, 2nd edition, McGraw-
Hill, New York, 1966.
2
s η
t axis s axis
mapped here T mapped here
T 2
2 2 1
3 1
t τ
−
T T −T T
2 2
4
−
T 4
2 3
−T
Figure 1: The domain of integration (gray regions) for the Fourier transform
of the autocorrelation Eq. (7): (left) for the original variables, t and s; (right)
for the transformed variables, η and τ , obtained by the change of variables
Eq. (8). Notice that the square region on the left is not only rotated (and
flipped about the t axis), but its area is increased by a factor of |J| = 2. The
circled numbers show where the sides of the square on the left are mapped
by the change of variables. The lines into which the t and s axes are mapped
are also shown.
where |J| is the absolute value of the Jacobian for the change of variables
Eq. (8) given by ∂f ∂f
∂t ∂s −1 1
J = ∂g ∂g = = −2 . (10)
1 1
∂t ∂s
To determine the limits of integration needed for the right hand side of Eq.
(9), we need to refer to Fig. 1, in which the domain of integration is plotted in
both the original (t, s) variables and the transformed (τ, η) variables. Since
we wish to integrate on η first, we hold τ fixed. For τ > 0, a vertical cut
through the diamond-shaped region in Fig. 1 (right) shows that −T + τ ≤
η ≤ T − τ , whereas for τ < 0 one finds that −T − τ ≤ η ≤ T + τ . Putting
3
this all together yields:
T T −|τ |
1
Z Z
E[XT XT∗ ] = Rx (τ ) e−2πif τ dη dτ
2 −T −(T −|τ |)
(11)
T
|τ |
Z
=T 1− Rx (τ ) e−2πif τ dτ .
−T T
Finally, dividing both sides of Eq. (11) by T and taking the limit as T → ∞
gives
Z T
1 ∗ |τ |
lim E[XT XT ] = lim 1− Rx (τ ) e−2πif τ dτ
T →∞ T T →∞ −T T
Z T
= lim Rx (τ ) e−2πif τ dτ
T →∞ −T (12)
Z ∞
= Rx (τ ) e−2πif τ dτ
−∞
= Sx (f ) .
Thus, in summary, the above demonstrates that
1
Sx (f ) = lim E[ |XT (f )|2 ] . (13)
T →∞ T
Recalling that XT (f ) has units SU/Hz (where SU stands for “signal units,”
i.e., whatever units the signal xT (t) has), it is clear that E[ |XT (f )|2 ] has
units (SU/Hz)2 . However, 1/T has units of Hz, so that Eq. (13) shows that
the PSD has units of (SU2 )/Hz. (Of course, the units can also be determined
by examining the definition of Eq. 1.)
Although it is not always literally true, in many cases the mean square of the
signal is proportional to the amount of power in the signal.4 The fact that
4
This comes primarily from the fact that, in electrical circuits, the power can be written
in terms of the voltage as V 2 /Z, or in terms of the current as I 2 Z, where Z is the circuit
impedance. Thus, for electrical signals, it precisely true that the mean square of the signal
will be proportional to the power. Be forewarned, however, that the mean square of the
scaled signal, expressed in terms of the actual measured variable (such as displacement or
acceleration), will not in general be equal to the average mechanical power in the structure
being measured.
4
Sx is therefore interpreted has having units of “power” per unit frequency
explains the name Power Spectral Density.
Given the definition Eq. (1), we also have the dual relationship
Z ∞
Rx (τ ) = Sx (f ) e2πif τ df . (14)
−∞
Finally, if we assume that x(t) is ergodic in the autocorrelation, that is, that
T
1
Z
2
Rx (τ ) = E[x(t) x(t + τ )] = lim x(t) x(t + τ ) dt , (17)
T →∞ T − T2
5
where the last equality holds for any sample function x(t), then Eq. (15) can
be rewritten as
Z T Z ∞
1 2
2
lim x(t) dt = Sx (f ) df . (18)
T →∞ T − T −∞
2
This last identity makes it clear that, given any two frequencies f1 and f2 ,
the quantity Z f2
Sx (f ) df (19)
f1
represents the portion of the average signal power contained in signal fre-
quencies between f1 and f2 , and hence Sx is indeed a “spectral density.”